Received-SPF: pass (google.com: domain of linux-kernel-owner@vger.kernel.org designates 23.128.96.36 as permitted sender) client-ip=23.128.96.36;
Date:   Wed, 18 Oct 2023 12:14:07 -0700
From:   Josh Poimboeuf <jpoimboe@kernel.org>
To:     Borislav Petkov <bp@alien8.de>
Cc:     Ingo Molnar <mingo@kernel.org>, linux-kernel@vger.kernel.org,
        linux-tip-commits@vger.kernel.org,
        David Kaplan <david.kaplan@amd.com>,
        "Peter Zijlstra (Intel)" <peterz@infradead.org>, x86@kernel.org,
        David Howells <dhowells@redhat.com>
Subject: Re: [tip: x86/bugs] x86/retpoline: Ensure default return thunk isn't
 used at runtime
Message-ID: <20231018191407.n4ys6vefbio7z3sn@treble>
References: <169713303534.3135.10558074245117750218.tip-bot2@tip-bot2>
 <20231018132352.GBZS/caGJ8Wk9kmTbg@fat_crate.local>
 <ZS/f8DeEIWhBtBeb@gmail.com>
 <20231018151245.GCZS/17QhDGe7q6K+w@fat_crate.local>
 <20231018155433.z4auwckr5s27wnig@treble>
 <20231018175531.GEZTAcE2p92U1AuVp1@fat_crate.local>
 <20231018181431.skre6i6vzrxsprck@treble>
 <20231018182223.GFZTAiX4BJ6FT8bHzz@fat_crate.local>
 <20231018183915.xwamzzqjf6gehaou@treble>
 <20231018184431.GGZTAnj8V/R54S5KK9@fat_crate.local>
MIME-Version: 1.0
Content-Type: text/plain; charset=utf-8
Content-Disposition: inline
In-Reply-To: <20231018184431.GGZTAnj8V/R54S5KK9@fat_crate.local>
Precedence: bulk

On Wed, Oct 18, 2023 at 08:44:31PM +0200, Borislav Petkov wrote:
> On Wed, Oct 18, 2023 at 11:39:15AM -0700, Josh Poimboeuf wrote:
> > And see my reply to that?  Not trying to be daft, I just didn't see how
> > your reply was responsive.
> > 
> > A single WARN_ONCE() has the benefit of not overflowing dmesg, while
> > also making the warning available to those looking at dmesg (or the
> > taint flag), as those who care should be.
> 
> A single WARN once is not enough as this is security-sensitive. Warns do
> get ignored.
> 
> > A rate-limited WARN() is problematic, as it overflows dmesg (and
> > possibly wrapping other logs), potentially obscuring other important
> > data.
> 
> This will hopefully make people look by screaming louder. But no
> guarantee. Not saying it is the right thing.
> 
> UDing without any output is not the right thing either.
> 
> All I'm saying is, there's no good solution for how to catch this and
> make people report it.
> 
> Make more sense?

There are a lot of warnings which could become security concerns.

By definition, a warning means something is seriously wrong.  If it's
ignored, all bets are off.  That's why we taint the kernel.

-- 
Josh